From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ido Schimmel Date: Sun, 7 Nov 2021 16:21:12 +0200 Subject: [Intel-wired-lan] [PATCH net-next 4/6] rtnetlink: Add support for SyncE recovered clock configuration In-Reply-To: References: <20211104081231.1982753-1-maciej.machnikowski@intel.com> <20211104081231.1982753-5-maciej.machnikowski@intel.com> <2d379392-a381-e60a-7658-5ac695c30df1@nvidia.com> Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: intel-wired-lan@osuosl.org List-ID: On Fri, Nov 05, 2021 at 12:17:19PM +0000, Machnikowski, Maciej wrote: > > > > -----Original Message----- > > From: Roopa Prabhu > > Sent: Thursday, November 4, 2021 7:25 PM > > To: Machnikowski, Maciej ; > > netdev at vger.kernel.org; intel-wired-lan at lists.osuosl.org > > Cc: richardcochran at gmail.com; abyagowi at fb.com; Nguyen, Anthony L > > ; davem at davemloft.net; kuba at kernel.org; > > linux-kselftest at vger.kernel.org; idosch at idosch.org; mkubecek at suse.cz; > > saeed at kernel.org; michael.chan at broadcom.com > > Subject: Re: [PATCH net-next 4/6] rtnetlink: Add support for SyncE recovered > > clock configuration > > > > > > On 11/4/21 1:12 AM, Maciej Machnikowski wrote: > > > Add support for RTNL messages for reading/configuring SyncE recovered > > > clocks. > > > The messages are: > > > RTM_GETRCLKRANGE: Reads the allowed pin index range for the > > recovered > > > clock outputs. This can be aligned to PHY outputs or > > > to EEC inputs, whichever is better for a given > > > application > > > > > > RTM_GETRCLKSTATE: Read the state of recovered pins that output > > recovered > > > clock from a given port. The message will contain the > > > number of assigned clocks (IFLA_RCLK_STATE_COUNT) and > > > a N pin inexes in IFLA_RCLK_STATE_OUT_IDX > > > > > > RTM_SETRCLKSTATE: Sets the redirection of the recovered clock for > > > a given pin > > > > > > Signed-off-by: Maciej Machnikowski > > > --- > > > > > > Can't we just use a single RTM msg with nested attributes ? > > > > With separate RTM msgtype for each syncE attribute we will end up > > bloating the RTM msg namespace. > > > > (these api's could also be in ethtool given its directly querying the > > drivers) > > I'm not a fan of merging those messages. The mergeable ones are > GETRCLKRANGE and GETCLKSTATE, but the get range function may be > result in a significantly longer call if the information about the underlying > HW require any HW calls. > They are currently grouped in 3 categories: > - reading the boundaries in GetRclkRange (we can later add more to it, like > configurable frequency limits) > - Reading current configuration > - setting the required configuration > > I don't plan adding any additional RTM msg types for those (and that's > the reason why I pulled them before EEC state which may have more > messages in the future) > > We also discussed ethtool way in the past RFCs, but this concept > is applicable to any transport layer so I'd rather not limit it to ethernet > devices (i.e. SONET, infiniband and others). I'm still not convinced that this doesn't belong in ethtool. I find it very weird to have a message called "Get Ethernet Equipment Clock State" in rtnetlink and not in ethtool. I also believe it was a mistake to add DCB to rtnetlink (for example, why PAUSE is configured via ethtool, but PFC via rtnetlink?)